Abstract
The journal occasionally publishes a series of articles on a topic of cutting-edge research that will advance our understanding of mental health and illness. The goal is to feature diverse models, methods, disciplines, and approaches. This series is on computational modeling, which encompasses the integration of mathematics, computers, and simulations to model complex systems, and in this case applied to psychiatric dysfunction, broadly conceived. Among the many strengths of computational modeling is the examination of multiple variations (simulations) of a set of variables, observed and inferred, and various parameters to provide better descriptions as well as to test and generate theories and explanations. One can begin by examining variables and their parameters to evaluate outcomes (e.g., clinical dysfunction, impairment in a specific domain or process), but also one can begin with an outcome to identify possible etiologies and causal paths. Computerization permits the evaluation of many (e.g., thousands) of variations to evaluate the outcomes that are likely in real systems.
The approach is generally familiar in two general ways. First, psychological research often builds and tests models (e.g., structural equation, goodness of fit, and regression models), and computational modeling is in keeping with that overall thrust. Yet, the scope of variables involved in complex systems, their interactions with each other, their multiple outcomes, and the interest and need to evaluate many different models are quite different. More powerful and sophisticated tools are used beyond those usually used in data analyses in psychological research. Computational modeling provides these tools but as important (if not moreso), it also provides a conceptual view about the models (e.g., interacting variables, changing parameters) that can better account for and describe phenomena we study.
Second, we are familiar with computational modeling in yet another way. Anyone who has followed The Weather Channel listens to the discussion of predicting the weather or an eminent storm or major weather event. Multiple computer simulations are used in which values of parameters are considered along with scores of variables and their dynamic relations. That is, computational models are used—multiple models, with simulations of what is likely to happen if this or that variable or estimated parameter changes (e.g., wind, speed and direction of the storm, water temperature). The scale of evaluation (variables, parameters), considering ongoing changes in the system, and the type of effects sought (beyond looking for main effects or interactions of one or a few variables) better capture the complexity and the dynamics involved in and causing the weather. In my own thinking, consider computational modeling as more like science’s version of The Weather Channel in the sense of testing whether a particular outcome will happen in light of multiple simulated variations of many variables and parameters.
Computational modeling is now applied to many areas in the natural, biological, and social sciences. Certainly in psychology the understanding of human functioning in general (affect, cognition, and behavior) and in specific areas (e.g., decision making and psychopathology) is influenced greatly by multiple variables operating in complex and interactive, dynamic, and reciprocal ways. Surely human behavior is at least as complex as the weather. Indeed, maybe computational modeling could help us understand some of our weathery colleagues, friends, and family members whose personalities and demeanors seem chilly, sunny, and gloomy, leaving aside those of us who live in fear that our preadolescent will be one of the few “storm and stress” adolescents we heard about. Computational models of clinical dysfunction represent an important advance and development that is not yet commonly discussed or presented in the mental health disciplines. Related advances in many areas of generating, evaluating, and assessing human functioning (e.g., assessment of neural processes, statistical analyses and modeling, computerization, “big data,” and real-time analyses of human functioning in daily life) permit model development and evaluation never available in the same way.
Our journal was extremely fortunate in recruiting Tiago V. Maia to serve as guest editor and develop a series of articles to illustrate advances in computational modeling as applied to psychiatry. Professor Maia’s own research has explored modeling and neural structures and function in relation to psychological processes (e.g., consciousness, emotional processing) and diverse disorders (e.g., obsessive-compulsive disorder, attention-deficit/hyperactivity disorder, Tourette’s syndrome). His work already has tackled conceptual issues in cognitive neuroscience, which is a primary venue for computational modeling, and he has developed programmatic research that is on the forefront using these methods. In light of his work in the area, I was delighted he agreed to guest edit this series. He also graciously agreed to go beyond editing and to provide a sample from his own research program.
We are at such an exciting time in science with advances converging in conceptual approaches, data sets, and measures to analyze phenomena at multiple levels (e.g., nanoscale to culture). Many “individual” influences need to be integrated into our models and at multiple levels (e.g., the microbiome, connectome, exposome), and they continue to point to increased complexity and variation in processes and their outcomes. Computers, math, and simulations reflect powerful ways of addressing both conceptual issues and actual predictions. The potential is hardly tapped, but already we have an active group of researchers making inroads, as reflected in the articles that follow.